Deep Learning Dimensional Measurement for Industrial Tolerance

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing dimensional measurement methods in industrial production face challenges with reduced repeatability and accuracy due to environmental and material inconsistencies, which affect the precision of positioning and dimensional data obtained through template matching and manual feature identification.

Innovation Solution

A dimensional measurement method and device utilizing deep learning that captures images of target objects with preset location precision, determines target regions, and processes them using pre-trained neural networks to obtain precise position information and dimensional data, improving the accuracy and repeatability of measurements.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If template matching is used for positioning, then the measurement process is simple, but the positioning accuracy and repeatability deteriorate due to environmental and material variations

Engineering Contradiction:

Improvesimplicity of measurement processVSAvoidpositioning accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent replaces traditional template matching algorithms with a deep learning-based neural network system. The neural network is trained to recognize and locate target features in images, substituting the mechanical/template-based approach with an intelligent system that can adapt to environmental variations while maintaining high positioning accuracy and repeatability

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the measurement approach from fixed template matching to a learned feature representation system. By training the neural network on diverse samples, the system learns robust feature parameters that remain stable across different environmental conditions, thereby improving measurement precision without sacrificing process simplicity

Inventive Principle:

Principle #35Parameter changes

2Ease of operation

If manual feature identification is used, then the measurement method is straightforward, but the dimensional data accuracy deteriorates due to inconsistencies in environmental and material conditions

Engineering Contradiction:

Improvestraightforwardness of measurement methodVSAvoiddimensional data accuracy

Core Design Contradiction:

Ease of operationVSManufacturing precision

Solution Approach 1:

The patent replaces manual feature identification with an automated deep learning system. The neural network automatically identifies and extracts relevant features from images, eliminating manual intervention while providing consistent and accurate dimensional measurements that are robust to environmental and material variations

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The neural network performs self-learning and adaptation through training on labeled data. The system automatically adjusts its feature extraction capabilities to handle different environmental conditions and material variations, providing accurate dimensional measurements without requiring manual tuning or intervention for each new condition

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS20240062358A1Dimensional Measurement Method Based on Deep Learning

Publication Date: 2024.02.22 UNITX INC

AI summary

The present disclosure provides a dimensional measurement method and device based on deep learning. The method includes capturing a target image of a target object, obtaining measurement data for the target image, and determining whether or not the target object is within a preset tolerance.